{"id":416204,"date":"2026-04-29T06:51:08","date_gmt":"2026-04-29T06:51:08","guid":{"rendered":"https:\/\/www.newsbeep.com\/il\/416204\/"},"modified":"2026-04-29T06:51:08","modified_gmt":"2026-04-29T06:51:08","slug":"openais-instructions-to-codex-have-a-weirdly-emphatic-no-creatures-policy","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/il\/416204\/","title":{"rendered":"OpenAI\u2019s Instructions to Codex Have a Weirdly Emphatic No-Creatures Policy"},"content":{"rendered":"

A document posted by OpenAI on Github<\/a> as part of the open-sourcing for Codex CLI, OpenAI\u2019s most recent flagship coding agent, contains what looks like the entire system prompt for GPT-5.5 in a coding context. And it seems to correct for the model\u2019s past addiction to talking about whimsical creatures both natural and supernatural.<\/p>\n

Here\u2019s the relevant section (emphasis added):<\/p>\n

\u201cprovide the highest-signal context instead of describing everything exhaustively.\\n- Tone of your final answer must match your personality.\\n- Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user\u2019s query.\u201d<\/p>\n

Evidently this point is so important, the developers bring it up again a bit later:<\/p>\n

\u201cFor example, never use platitudes like \\\u201dI will do rather than \\\u2019, \\\u2019I will do , not \\\u2019.\\n- Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user\u2019s query.\u201d<\/p>\n

It\u2019s not clear why this matters so much. If these were, for instance, canary words<\/a> inserted into the system prompt as a way to monitor for prompt injection attacks, they would probably be more random, rather than seemingly a whole category of animals.<\/p>\n

OpenAI did not reply to a request for comment on Tuesday night.<\/p>\n

A Google employee named Barron Roth posted what appears to be a search of his chat logs with some of his GPT-5.5-powered Openclaw agents, showing that at least one had a history of inserting the word \u201cgoblin\u201d into messages to the user multiple times in a single day. To my admittedly untrained eye, GPT-5.5 seems to be using it in place of a word like \u201cthingy.\u201d<\/p>\n

this explains gpt5.5\u2019s goblin adoration in openclaw and why @pashmerepat<\/a> is making us move to the codex harness \ud83e\udd23\ud83e\udd23\ud83e\udd23 pic.twitter.com\/N20UhQDOK5<\/a><\/p>\n

\u2014 Barron Roth (@iamBarronRoth) April 28, 2026<\/a><\/p>\n

Nick Pash, who works on Codex at OpenAI, partially confirmed that Roth had hit on the nature of the problem, writing to him<\/a> on X \u201cthis is indeed one of the reasons.\u201d<\/p>\n

It looks like other X users also noticed that Codex had taken the goblin talk a little too far<\/a>.<\/p>\n

On Tuesday, this goblin issue became a meme, with users suggesting<\/a> that some sort of \u201cGoblin Mode,\u201d could be toggled on and off.<\/p>\n

Pash eventually added his own post about Goblin Mode:<\/p>\n

Alright fine. pic.twitter.com\/6RV1SHx78C<\/a><\/p>\n

\u2014 pash (@pashmerepat) April 28, 2026<\/a><\/p>\n

If this is giving you flashbacks to a year ago when OpenAI was all about Studio Ghibli memes<\/a>, you\u2019re not alone. Some dared to assert<\/a> that OpenAI might be doing all this goblin stuff for attention at what is, after all, a somewhat troubled moment for the company<\/a>.<\/p>\n

But according to another post by Pash<\/a>, \u201cit really isn\u2019t a marketing gimmick.\u201d<\/p>\n